reagent.ope.datasets package

Submodules

reagent.ope.datasets.logged_dataset module

class reagent.ope.datasets.logged_dataset.BanditsDataset

Bases: abc.ABC

Base class for logged, aka behavior, dataset

abstract property actions: torch.Tensor
Returns

all actions in the dataset as numpy array

abstract property features: torch.Tensor
Returns

all features in the dataset as numpy array

abstract property num_actions: int
Returns

number of total possible actions

abstract property num_features: int
Returns

number of features

abstract property rewards: torch.Tensor
Returns

all rewards in the dataset as numpy array

Module contents